Texture Image Segmentation Using Spectral Histogram and Skeleton Extracting

This paper presents a texture image segmentation algorithm using spectral histogram and skeleton extracting. No need of selecting seed pixels or specifying or deciding the number of regions is its remarkable characteristic. Based on a local spatial/frequency representation, spectral histogram consists of marginal distributions of responses of a bank of filters and encodes implicitly the local structure of images through the filtering stage and the global appearance through the histogram stage. The similarity between two spectral histograms is measured using chi²-statistic. According to the similarity among spectral histograms, the initial binary segmentation image can be obtained. Adopting skeleton extracting algorithm based on mathematical morphology, the final segmentation result can be obtained. Experiments on Brodatz textures give satisfactory results.